结合自动化成像设备与深度学习算法的数字化免疫分析系统检测牛奶中的恩诺沙星残留

Digital immunoassay system integrating automated imaging equipment with deep learning algorithms for detecting enrofloxacin residues

中文关键词:  恩诺沙星，AIDS，聚苯乙烯微球，深度学习

英文关键词:enrofloxacin  digital immunoassay system  polystyrene microspheres  deep learning

中文摘要:

      目的 开发一种智能化、高灵敏的恩诺沙星(enrofloxacin, ENR)快速检测平台, 以解决传统检测方法成本高、耗时长, 以及现有数字免疫分析技术依赖人工操作、识别准确性不足等问题。方法 基于自动化成像设备与深度学习算法, 构建以聚苯乙烯微球为信号载体的数字化免疫分析系统, 结合免疫反应与磁性分离技术实现定量检测。该系统集成高精度自动显微位移平台, 采用Faster-RCNN算法实现微球快速识别, 并引入NAFNet图像增强算法以提升成像质量。结果 该系统能够实现ENR的高灵敏度检测(检出限19.35 pg/mL, 线性范围0.5~100.0 ng/mL), 测试回收率为104%~111%, 且在检测效率、自动化程度和抗干扰能力上均取得了显著突破。结论 。本研究构建了基于深度学习的自动化免疫检测平台, 可实现微球图像稳定采集与精准识别。深度学习算法计数准确可靠, 搭建的免疫检测体系线性范围宽、检出限低、灵敏度高, 在牛奶样品中检测效果良好, 适用于食品中恩诺沙星快速自动化检测。

英文摘要:

      Objective To develop an intelligent, highly sensitive rapid detection platform for enrofloxacin (ENR) to address the issues of high cost and lengthy processing times associated with traditional detection methods, as well as the reliance on manual operation and insufficient identification accuracy of existing digital immunoassay technologies. Methods Based on automated imaging equipment and deep learning algorithms, a digital immunoassay system using polystyrene microspheres as signal carriers was constructed, combining immunoreaction and magnetic separation technologies to achieve quantitative detection. The system integrated a high-precision automated microdisplacement platform, adopted the Faster-RCNN algorithm to achieve rapid recognition of microspheres, and introduced the NAFNet image enhancement algorithm to improve the imaging quality. Results The platform was capable of achieving high sensitivity detection of ENR (limit of detection 19.35 pg/ml, linear range 0.5–100.0 ng/mL), with test recoveries ranging from 104%–111%, and had made significant breakthroughs in detection efficiency, automation and anti-interference capability. Conclusion The study develops a deep learning-based automated immunoassay platform capable of stable image acquisition and accurate recognition of microspheres. The deep learning algorithm provides accurate and reliable counting, and the immunoassay system establishes features a wide linear range, low detection limit and high sensitivity. It performs well in the detection of milk samples and is suitable for the rapid automated detection of enrofloxacin in food.